GRWG Web Meeting 2013-07-03
GSICS Web Meeting on Rayleigh Scattering
- Bertrand Fougnie (CNES) Introductory Presentation
- Bertrand Fougnie (CNES) Draft ATBD for GEO-LEO inter-calibration by Rayleigh Scattering
- All - Discussion on way forward
Guest Chair: Dave Doelling (NASA)
CNES: Bertrand Fougnie
EUMETSAT: Tim Hewison, Bartolomeo Viticchie
JMA: Masay Takahashi, Keita Hosaka, Kenji Date
NASA: Dave Doelling, Bob Iacovazzi
NOAA: Fangfang Yu
UAH: Denis Buechler
Betrand Fougnie (CNES) reviewed the Rayleigh scattering inter-calibration technique as outlined in the draft ATBD. This technique was derived from LEO sensors but was also applied to SEVIRI geostationary imager.
1) Clear-sky selection: He first outlined clear-sky ocean selection process, which included, ocean domains, angular and sunglint constraints, surface wind speed threshold, and a cloud, white cap and aerosol mask. The key selection process is the turbidity rejection, which captures any residual clouds, aerosols, and white caps, and is based on the 0.86µm channel. The 0.86µm channel threshold filters ~80% of potential clear-sky pixels.
2) Compute predicted clear-sky radiance: The predicted 0.65µm and shorter wavelength band reflectance are computed using the 0.86 µm well calibrated reflectance and contributions from gaseous transmittance and surface and aerosol reflectance, which requires reanalysis input of surface pressure and wind speed, O3
0 concentration, aerosol AOD. Other inputs are based on climatology. He suggests using the 6S model. Due to surface reflectance model limitations, a view and solar angle limit of 60° was suggested.
3) Error Budget: The error budget is a function of wavelength and therefor the band spectral response function must be known to 1 nm. For the 0.65µm wavelength (red channel) the 0.86µm calibration is most important, whereas for the blue channel the ocean reflectance is most important. Due to the seasonal dependencies of the ocean surface reflectivity, multiple clear-sky domains are needed to reduce the uncertainty. Need to know the polarization of the sensor being calibrated.
Bertrand mentioned a paper
he presented at SPIE in 2010, which discussed the regional and seasonal variations of the maritime relectivity. This implies that multiple ocean sites are needed in both Northern and Southern Hemispheres to compensate for these variations.
Challenges for GEO calibration application:
Limited clear-sky ocean domains with view angles < 60° and most sites are in constant backscatter conditions. Most operational GEO sensors do not have a 0.86µm channel, which is needed for both clear-sky detection and model radiance computation. Although a SWIR channel could, in principle, be used instead for this purpose, most current GEO imagers also lack this. It was suggested to use the MODIS 0.86µm turbidity test, calibration and aerosol AOD, albeit with a different view and azimuth geometry than the GEO sensor to be calibrated. However, it was recognized that this would lead to increased uncertainty in that component of the error budget. Perhaps a histogram technique could be used to determine the valid predicted clear-sky pixel-level radiances. It still needs to be investigated if the gain and offset can be determined from this method, in order assess instrument stray light.
: Betrand Fougnie (CNES) to revisit the rayliegh scattering error budget in ATBD Table 2 and report results at the next GSICS annual meeting.
: Betrand Fougnie (CNES) to provide 3 months of SEVIRI Rayleigh scattering inputs and predicted radiance, which is on the internal SADE database to the GSICS archive. Contact Aleksandar Jelenak.
: Sebastien Wagner (EUMETSAT) to compare SEVIRI predicted radiances from CNES and the EUMETSAT operational SEVIRI Solar Channel Calibration system and report results at the next GSICS annual meeting.
: Masaya Takahashi (JMA) to evaluate the uncertainties in JMAs current radiative transfer technique using MODIS aerosols and other inputs to predict MTSAT visible clear-sky radiances, following Bertrands example.